Identification of Cancer Hub Gene Signatures Associated with Immune

Home , Ding (surname), LIN9, Peng (surname), RAB31, RHOQ, TNPO2

cancers

Article Identiﬁcation of Cancer Hub Gene Signatures Associated with Immune-Suppressive Tumor Microenvironment and Ovatodiolide as a Potential Cancer Immunotherapeutic Agent

Jia-Hong Chen 1,2, Alexander T. H. Wu 3,4,5,6 , Bashir Lawal 7,8 , David T. W. Tzeng 9, Jih-Chin Lee 10, Ching-Liang Ho 2 and Tsu-Yi Chao 1,2,11,12,*

1 Graduate Institute of Clinical Medicine, College of Medicine, Taipei Medical University, Taipei 11031, Taiwan; [email protected] 2 Division of Hematology/Oncology, Department of Medicine, Tri-Service General Hospital, National Defence Medical Center, Taipei City 114, Taiwan; [email protected] 3 The PhD Program of Translational Medicine, College of Science and Technology, Taipei Medical University, Taipei 11031, Taiwan; [email protected] 4 Clinical Research Center, Taipei Medical University Hospital, Taipei Medical University, Taipei 11031, Taiwan 5 Graduate Institute of Medical Sciences, National Defense Medical Center, Taipei 11490, Taiwan 6 Taipei Heart Institute (THI), Taipei Medical University, Taipei 11031, Taiwan 7 PhD Program for Cancer Molecular Biology and Drug Discovery, College of Medical Science and Technology, Taipei Medical University and Academia Sinica, Taipei 11031, Taiwan; [email protected] 8 Graduate Institute for Cancer Biology & Drug Discovery, College of Medical Science and Technology, Taipei, Medical University, Taipei 11031, Taiwan 9 School of Life Sciences, The Chinese University of Hong Kong, Hong Kong; [email protected] Citation: Chen, J.-H.; Wu, A.T..H.; 10 Department of Otolaryngology-Head and Neck Surgery, Tri-Service General Hospital, National Defense Lawal, B.; Tzeng, D.T.W.; Lee, J.-C.; Medical Center, 325 Cheng-Kung Road Section 2, Taipei City 114, Taiwan; [email protected] Ho, C.-L.; Chao, T.-Y. Identification of 11 Division of Hematology and Oncology, Department of Internal Medicine, Taipei Medical University-Shuang Cancer Hub Gene Signatures Ho Hospital, New Taipei City 235, Taiwan Associated with Immune-Suppressive 12 Taipei Cancer Center, Taipei Medical University, Taipei City 11031, Taiwan Tumor Microenvironment and * Correspondence: [email protected] Ovatodiolide as a Potential Cancer Immunotherapeutic Agent. Cancers Simple Summary: In order to identify common genes associated with the pathology of multiple 2021, 13, 3847. https://doi.org/ cancers, we integrated differential expressed gene (DEGs) from datasets of six cancers (liver, lung 10.3390/cancers13153847 colorectal, gastric, prostate, and breast cancers) and identified six DEGs common to the six cancers. We conducted enrichment analysis and our results suggested that the DEGs are involved in the Academic Editors: Alan Hutson and tumorigenic properties, including distant metastases, treatment failure, and survival prognosis. Song Liu Notably, our results suggested high frequencies of genetic and epigenetic alterations of the DEGs in association with tumor staging, immune evasion, poor prognosis, and therapy resistance. Transla- Received: 9 June 2021 Accepted: 27 July 2021 tionally, we intended to identify a drug candidate with the potential for targeting the DEGs. Using a Published: 30 July 2021 molecular docking platform, we estimated that ovatodiolide, a bioactive anti-cancer phytochemical, has high binding affinities to the binding pockets of the hub genes and thus could serve as a potential

Publisher’s Note: MDPI stays neutral drug candidate for targeting the DEGs. with regard to jurisdictional claims in published maps and institutional afﬁl- Abstract: Despite the signiﬁcant advancement in therapeutic strategies, breast, colorectal, gastric, iations. lung, liver, and prostate cancers remain the most prevalent cancers in terms of incidence and mortality worldwide. The major causes ascribed to these burdens are lack of early diagnosis, high metastatic tendency, and drug resistance. Therefore, exploring reliable early diagnostic and prognostic biomarkers universal to most cancer types is a clinical emergency. Consequently, in the present Copyright: © 2021 by the authors. study, the differentially expressed genes (DEGs) from the publicly available microarray datasets of Licensee MDPI, Basel, Switzerland. six cancer types (liver, lung colorectal, gastric, prostate, and breast cancers), termed hub cancers, This article is an open access article were analyzed to identify the universal DEGs, termed hub genes. Gene set enrichment analysis distributed under the terms and (GSEA) and KEGG mapping of the hub genes suggested their crucial involvement in the tumorigenic conditions of the Creative Commons properties, including distant metastases, treatment failure, and survival prognosis. Notably, our Attribution (CC BY) license (https:// results suggested high frequencies of genetic and epigenetic alterations of the DEGs in association creativecommons.org/licenses/by/ with tumor staging, immune evasion, poor prognosis, and therapy resistance. Translationally, we 4.0/).

Cancers 2021, 13, 3847. https://doi.org/10.3390/cancers13153847 https://www.mdpi.com/journal/cancers Cancers 2021, 13, 3847 2 of 20

intended to identify a drug candidate with the potential for targeting the hub genes. Using a molecular docking platform, we estimated that ovatodiolide, a bioactive anti-cancer phytochemical, has high binding afﬁnities to the binding pockets of the hub genes. Collectively, our results suggested that the hub genes were associated with establishing an immune-suppressive tumor microenvironment favorable for disease progression and promising biomarkers for the early diagnosis and prognosis in multiple cancer types and could serve as potential druggable targets for ovatodiolide.

Keywords: cancer hub; tumor microenvironments; onco-immune proﬁling; differentially expressed genes; ovatodiolide; cancer immunotherapy

1. Introduction The current global burden of cancer was estimated to be 19.3 million cases in 2020 and is expected to be 28.4 million cases in 2040 [1], a 47% rise from 2020. In line with the previous trend [2,3], the 2020 global cancer statistic report of cancer incidences and mortality indicate that breast (11.7% and 6.9%), lung (11.4% and 18%), colorectal (10.0% and 9.4%), prostate (7.3% and 3.8%), stomach (5.6% and 7.7%), and liver (4.7% and 8.3%) cancers are the most commonly diagnosed and the leading cause of cancer mortality globally [1]. With the exception of prostate and breast cancers, which are sex-specific, the prevalence and mortality rates of these cancers combined were higher in men than in women, at 45.5% and 42.1%, respectively. These six cancers, which constituted 50.7% and 54.1% of the total global cases and cancer deaths, respectively, are the primary focus of the current study and are collectively referred to as hub cancers hereafter. Therefore, it is crucial to identify the promising early diagnostic and prognostic biomarkers that may assist in elucidating the underlying molecular mechanisms of these cancers and simultaneously improve the clinical therapeutics [4]. Microarray technology and bioinformatics analysis have become a promising and valuable tool for screening significant genetic or epigenetic variations that occur during carcinogenesis and understanding the pathogenesis of the diseases for effective diagnosis, prognosis and planning adequate therapeutic strategies [5–8]. However, the outcome from most of these findings has not yielded significant translational success in predicting reliable universal biomarkers for the diagnosis and prognosis of the hub cancers. Thus, independent analysis of microarray and RNAseq data from different cancer datasets with subsequent integration of differentially expressed genes (DEGs) across the cancer types may help identify the universal markers concordant across multiple cancer studies with increased confidence and translational relevance to the clinics. In addition to the lack of reliable and early diagnostic biomarkers, the resistance to chemotherapy and adverse effects associated with chemotherapeutic agents currently use in clinics are jointly responsible for the poor prognosis of the hub cancer patients [9,10]. Therefore, there is an urgent need to develop new, affordable, effective and safer anticancer drugs [11]. Ovatodiolide, is a phytochemical isolated from Anisomeles indica (L.) Kuntze [12]. This plant originates from Taiwan and is commonly known as “Fang Feng Cao” by the traditional Taiwanese herbalist where it is commonly used as an oral remedy for stom- achache, swelling, abdominal pain, hypertension, arthritis, immunodeficiency disease, hepatic diseases, hemorrhoids, and arthritis [13]. The plant has been reported for various biological activities, including analgesic, anti-inflammatory, antimicrobial, hypotensive, hepatoprotective, and anti-proliferative activities [14]. Ovatodiolide has been identified as the major bioactive compound responsible for the biological activities of this plant and has been previously isolated and evaluated for anti-cancer activities against several human cancer cell lines [12]. In our previous studies, we found that ovatodiolide sensitizes aggressive human cancer cells to chemotherapy and ameliorates the cancer stemness phenotype and chemotherapy-associated toxicity in an in vitro or/and in vivo models of breast cancer [15], glioblastoma [16], oral squamous cell carcinoma [17,18], colon cancer [19], and nasopha- Cancers 2021, 13, 3847 3 of 20

ryngeal carcinoma [20]. Studies elsewhere have also reported ovatodiolide for significant anti-cancer activities against hepatic cancer stem cells [21], renal [22], and oral [23] cancers. Our earlier mechanistic study indicated that inhibition of numerous oncogenic molecules and signaling pathways such as Wnt/β-catenin, Hippo/YAP1, PI3K/mTOR, exosomal Mir-21/STAT3/β-Catenin, JAK2/STAT3/JARID1B, tumor necrosis factor (TNF)-α, nuclear factor (NF)-κB, matrix metalloproteinases (MMPs), and FLICE inhibitory protein (FLIP) were associated with the anti-cancer activity of this bioactive phytochemical [15,17,18,20,22]. However, the therapeutic potential of ovatodiolide against the hub genes identified in the current study via bioinformatics integrations of DEGs from the hub cancers has never been explored. Herein, our molecular docking analysis suggested that ovatodiolide docked well into the binding sites of these hub genes with estimated higher binding preferences for IRAK3, SEC168, and TNPO2; this suggested that ovatodiolide could be a drug candidate for targeting these oncogenic hub genes. In summary, the identified hub genes may provide novel insights on the early diagnosis and prognosis of hub cancers by serving as promising biomarkers/druggable targets for ovatodiolide.

2. Methods 2.1. Collection of Microarray of Cancer and Normal Samples The microarray datasets of six cancer types (lung, liver, colorectal, prostate, gastric, and breast cancers) were collected from the NCBI Gene Expression Omnibus (GEO), a public functional genomics data repository of high throughput gene expression (http://www.ncbi.nlm.nih.gov/geo/, accessed on 11 May 2021, Figure1). The dataset containing primary or metastatic cancer tissues (tumor samples) and normal human samples (normal counterparts) were included (Table1). Identiﬁcation of DEGs was performed using the LIMMA R package. The Benjamini–Hochberg correction method was used for p-value adjustment to false discovery rate (FDR). FDR < 0.05 and |logFC| > 1.5 was set as a cutoff point for DEGs selection. Online tool Multiple List Comparator (https://www.molbiotools.com/listcompare.html, accessed on 19 May 2021) was used to visualize the intersected DEGs and generate a Venn diagram for the visualization of the overlapping DEGs. In addition, we collected TCGA data of mRNA expression levels, copy number variation (CNV), single nucleotide variation (SNV), and methylation levels of the six hub genes in LUAD, BRCA, STAD, COAD, and LICH patients as well as the survival duration of these patients from the National Cancer Institute (NCI) Genomic Data Commons (GDC) (https://gdc.cancer.gov/, accessed on 27 May 2021). The distribution of the sample collection is presented in Table2. All analyses of GDC data were conducted using the GSCLite web package.

Table 1. Characteristics of datasets used for the identiﬁcation of the differentially expressed genes (DEGs).

Cancers Dataset Platform Tumor Normal Breast GSE103512 GPL13158 [HT_HG-U133_Plus_PM] Affymetrix HT HG-U133+ PM Array Plate 65 10 Colorectal GSE103512 GPL13158 [HT_HG-U133_Plus_PM] Affymetrix HT HG-U133+ PM Array Plate 57 12 Lung GSE103512 GPL13158 [HT_HG-U133_Plus_PM] Affymetrix HT HG-U133+ PM Array Plate 60 9 Prostate GSE72220 GPL5175 [HuEx-1_0-st] Affymetrix Human Exon 1.0 ST Array 57 90 Liver GSE19665 GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array 10 10 Stomach GSE79973 GPL570 HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array 10 10

Table 2. Distribution of data collection from the NCI Genomic Data Commons.

Cancers mRNA Expression Survival Stage CNV SNV Methylation LUAD 578 481 473 517 567 528 BRCA 1220 1037 1034 1081 1026 816 STAD 452 373 341 442 439 416 PRAD 552 354 0 493 498 576 COAD 331 442 433 452 407 361 LIHC 426 197 176 371 365 453 CNV: copy number variation, SNV: single nucleotide variation. Cancers 2021, 13, 3847 4 of 20 Cancers 2021, 13, x 4 of 23

FigureFigure 1. The 1. The ﬂow flow chart chart of of the the study study design design for for identifying identifying the key DEGs DEGs in in the the pathology pathology of ofthe the six six cancers cancers and and predicting predicting theirtheir potential potential druggability druggability for for ovatodiolide. ovatodiolide.

Table 1. Characteristics of datasets used for the identification of the differentially expressed genes (DEGs). 2.2. Differential Expression Analysis of the Hub Genes between Tumor Stage and Cancers Dataset Molecular Subtypes Platform Tumor Normal Breast GSE103512 GPL13158In order [HT_HG-U133_Plus_PM] to predict the clinical Affymetrix relevance HT ofHG-U133+ the hub PM genes, Array we Plate analyzed 65 the differential 10 Colorectal GSE103512 GPL13158expression [HT_HG-U133_Plus_PM] levels of the hub genes Affymetrix between HT HG-U133+ tumor stagePM Array and Plate molecular 57 subtypes. 12 To Lung GSE103512 GPL13158make the [HT_HG-U133_Plus_PM] subtypes analysis feasible, Affymetrix the HT number HG-U133+ of subgroupsPM Array Plate of subtypes 60 must 9 have Prostate GSE72220 GPL5175 [HuEx-1_0-st] Affymetrix Human Exon 1.0 ST Array 57 90 at least 10 samples. Only lung, stomach, and breast cancer met this criterion and hence Liver GSE19665 GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array 10 10 Stomach GSE79973 GPL570were included HG-U133_Plu for thes_2] subtype Affymetrix analysis. Human Genome The RNA-Seq U133 Plus by2.0 Array Expectation-Maximization 10 10 (RSEM) normalized expression values were used for the differential expression levels of Tablethe 2. hubDistribution genes betweenof data collection the molecular from the subtypesNCI Genomic of gastricData Commons. cancer (Epstein–Barr virus (EBV), microsatellite instability (MSI), genomically stable (GS) and chromosomally unstable (CIN)), mRNA Cancers breast cancer (basal,Survival Her2, LumA,Stage and LumB)CNV and lungSNV cancer (subtypes Methylation 1–6). Analysis of Expression variance (ANOVA) t-test was used for the statistical analysis. A p-value of less than 0.05 LUAD 578 481 473 517 567 528 BRCA 1220was considered signiﬁcant. 1037 1034 1081 1026 816 STAD 452 373 341 442 439 416 2.3. Prognosis Analysis of the Hub Genes PRAD 552 354 0 493 498 576 COAD 331 We queried the 442 associations433 between the452 gene expression407 profile and361 prognosis of the LIHC 426cohorts by integrating 197 tumor gene176 expression levels371 of the hub365 genes and overall453 survival data fromCNV: the copy patients number of eachvariation, of the SNV: hub single cancers nucleotide using thevariation. PREdiction of Clinical Outcomes from Genomic profiles (PRECOG) server (https://precog.stanford.edu/index.php, accessed on 292.2. May Differential 2021). Similarly, Expression we Analysis queried of the the survivalHub Genes differences between Tumor between Stage the and mRNA Molecular expression levelsSubtypes of each hub gene across the combined cohorts of the hub cancers using the Gene ExpressionIn order Profiling to predict Interactive the clinical Analysis relevance (GEPIA) of the hub webtool. genes, Kaplan–Meirwe analyzed the plots differential were used toexpression visualized levels the survival of the hub differences genes between between tumor hub stage cancer and cohorts molecular with subtypes. high and To lowmake gene expressionthe subtypes levels. analysis feasible, the number of subgroups of subtypes must have at least 10 samples. Only lung, stomach, and breast cancer met this criterion and hence were in- 2.4. Functional Enrichment Analysis of the Hub Genes Online databases (https://maayanlab.cloud/Enrichr/, accessed on 29 May 2021) were used for the gene ontology function and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analyses of the DEGs. The GO and KEGG terms with Cancers 2021, 13, 3847 5 of 20

FDR < 0.05 were regarded as signiﬁcant functions and pathways and were visualized using the R package cluster Proﬁler.

2.5. Gene–Gene and Protein–Protein Interaction Network Construction Analysis Network construction for gene—gene interaction (GGI) and protein–protein interaction (PPI) of the hub genes were performed via the GENEMANIA (https://genemania.org/, accessed on 1 June 2021) and the Search Tool for the Retrieval of Interacting Genes (STRING; http://string.embl.de/, accessed on 23 May 2021), respectively. The PPIs of the DEGs were constructed with a conﬁdence score of 0.70.

2.6. Analysis of Single Nucleotide Variation (SNV) and Copy Number Variation (CNV) of the Hub Genes We collected SNV data of seven variant types of effective mutations (Missense_Mutation, Nonsense_Mutation, Frame_Shift_Ins, Splice_Site, Frame_Shift_Del, In_Frame_Del, In_Frame_Ins) of lung, breast, colon prostate, gastric, and liver cancer from the NCI Ge- nomic Data Commons (https://gdc.cancer.gov/, accessed on 29 May 2021), and analyzed the frequencies and occurrence of these SNV in the hub genes across the six hub cancers via the GSCALite (http://bioinfo.life.hust.edu.cn/web/GSCALite/, accessed on 29 May 2021) server [24]. SNV percentage of each gene’s coding region was calculated by: Num of Mutated Sample/Num of Cancer Sample. Mutation Annotation Format (MAF) tools were used for data visualization and to generate the waterfall plot [25]. The CNV data were processed with GISTICS2.0 [26]. The frequency of four types of CNV—Hete Amp: heterozygous amplification (CNV = 1), Hete Del: heterozygous deletion, (CNV = −1), Homo Amp: homozygous amplification (CNV = 2), and Homo Del: homozygous deletion (CNV = −2)—in the six hub genes across the six cancer types were visualized. The SNV or CNV data and clinical overall survival data were combined and analyzed for survival differences between mutated and non-mutated genes (SNV) and between CNV using R package survival. A log-rank test [27] was also performed to compare the distributions of groups while Cox regression [28] was performed to estimate the hazards. A p-value < 0.05 was considered as significant.

2.7. Methylation Analysis of the Hub Genes We used Pearson’s product–moment correlation coefficient and followed a t-distribution to analyze the correlation between the mRNA expression and methylation levels of the hub genes across the six cancer types. The p-value was adjusted by FDR and genes with FDR ≤ 0.05 were considered as significant. In addition, we collected the clinical overall survival data of the cohorts and analyzed the survival differences between hyper and hypo- methylation levels of the hub genes across the six cancers. A log-rank test was also performed to compare the distributions of two groups while Cox regression analysis was conducted to estimate the hazards (risk of death). A p-value < 0.05 was considered as significant.

2.8. Analysis of the Hub Genes Expressions Correlation with Gene Expression Proﬁles Suggestive of Immune and Immuno-Suppressive Cell Inﬁltrations Weused the Tumor Immune Estimation Resource (TIMER2.0) (http://timer.cistrome.org/, accessed on 21 May 2021) [29] to analyze the hub gene expression correlations with the levels of gene expression suggestive of immune infiltrations (B Cell, CD8+ T Cell, CD4+ T Cell, macrophage, neutrophil, and dendritic cells) in the six cancer types. We also analyzed the hub gene expression correlations with the gene expression profiles suggestive of immunosuppressive cells infiltration. Four immunosuppressive cells that are known to promote T cell exclusion—vis myeloid-derived suppressor cell (MDSCs), cancer-associated fibroblast (CAF), tumor-associated macrophages (M2-TAM), and regulatory T cell (Treg)—were included. The correlation analysis was conducted using the purity-corrected partial Spearman’s rho value and statistical significance (p < 0.05). We used GraphPad Prism Software (version 8.0.0 for Windows) for data visualization. Heat maps were used to visualize the correlations between the hub genes expression levels and the gene expression profiles suggestive of immune/immunosuppressive cell infiltrations across the six cancers. In addition, we used the Cancers 2021, 13, 3847 6 of 20

QUERY module of the TIDE algorithm to evaluate the correlations between the hub gene expression and levels of gene expression suggestive of T cell exclusion and dysfunctional T cell phenotypes of the six hub cancers [30].

2.9. Gene Prioritization Analysis of The Hub Genes We access the gene prioritization of the hub genes across two parameters, namely the response to ICB therapy and gene knockout phenotype in CRISPR screens. The z-score in the Cox-PH regression was used to evaluate the effect of the gene on the expression on patient survival in ICB treatment cohorts. The normalized logFC in CRISPR screens was employed to evaluate the effect of gene knockout on lymphocyte-induced tumor death in cancer models [30].

2.10. Drug Response and Sensitivity Analysis of the Hub Genes We used the Spearman correlation analysis to explore the correlation between mRNA expression levels of the hub genes and IC50 concentrations of small molecules against cancer cell lines in the Therapeutics Response Portal (CTRP) database. In addition, we also analyzed the correlation of the expression levels of the hub genes with sensitivity to chemotherapy of the cancer patient using the ROC Plotter (http://www.rocplot.org/, accessed on 28 May 2021) algorithm, a transcriptome-based tool for predictive biomarkers by linking gene expression and response to therapy in cancer patients [31].

2.11. Molecular Docking Studies The molecular docking study of the hub genes with ovatodiolide was conducted using AutoDock VINA (version 0.8) [32] software according to the protocols described in previous studies [33,34]. The three-dimensional (3D) structure of the hub genes RAB31 (PDB:2fg5), IRAK3 (PDB:6ruu), OBSCN (PDB:2e01), LIN9 (PDB:6c48), TNPO2 (PDB:2z5j), and SEC16B (PDB:3mzk) were downloaded from the protein data bank (PDB). The 3D structure of the ovatodiolide was obtained in the Sybyl mol2 format using the Avogadro molecular builder and visualization tool version 1.XX [35] before subsequently being transformed into the protein data bank (PDB) format using the PyMOL Molecular Graphics System, version 1.2r3pre. The PDB ﬁles of the crystal structures of the receptors (hub genes) were transformed to pdbqt format by using AutoDock VINA (version 0.8) [32]. The removal of water molecules and the addition of Kolman charges and hydrogen atoms were carried out before docking simulation [36]. Visualization and analysis of the docked complex were performed using the PyMOL software and Discovery studio visualizer (version 19.1.0.18287, BIOVIA, San Diego, CA, USA) [37].

3. Results 3.1. RAB31/IRAK3/OBSCN/LIN9/TNPO2/SEC16B Are Hub Genes Associated with the Development of Breast, Lung, Colorectal, Liver, Prostate, and Stomach Cancers (Hub Cancers) The detailed information for the GEO datasets for each of the six cancer types is shown in Table1 while the distribution of DEGs in the six cancer types analyzed are represented by the volcano plots (Figure2A). Overall, we identiﬁed 3083 DEGs in colorectal (1545 up-regulated and 1538 downregulated), 2032 DEGs in breast (1070 up-regulated and 962 down-regulated), 4825 DEGs in prostate (2869 up-regulated and 1959 down-regulated), 3966 DEGs in lung (1500 up-regulated and 2466 down-regulated), 1589 DEGs in liver (947 up-regulated and 642 downregulated), and 2460 DEGs in gastric cancer (1499 up- regulated and 961 downregulated) (Figure2B,C). We further identiﬁed six overlapping up-regulated DEGs, RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B, while four down- regulated genes were obtained after the integrated analysis of six GEO datasets (Figure2D). Cancers 2021, 13, 3847 7 of 20 Cancers 2021, 13, x 8 of 23

Figure 2. Identiﬁcation of differentially expressed genes (DEG) between the cancer patients and healthy controls. Figure 2. Identification of differentially expressed genes (DEG) between the cancer patients and healthy controls. (A) Vol- (A) Volcano plot showing the distribution of DEGs between the cancer patients and healthy controls. The top DEGs cano plot showing the distribution of DEGs between the cancer patients and healthy controls. The top DEGs are repre- are representedsented satisfying satisfying the criteria the criteria of logFC of value logFC and value p < 0.05. and Thep < color 0.05. indicates The color high indicates expression high (red) expression and low expression (red) and low expression(blue). (blue). (B) The ( BVenn) The diagrams Venn diagrams of the overlapping of the overlapping up-regulated up-regulated and (C) down-regulated and (C) down-regulated DEGs in the six DEGs cancers. in the(D) sixHeat cancers. (D) Heatshowing showing the logFC the logFC distribution distribution of the of10 theoverlapping 10 overlapping DEGs (6 DEGs up-regulated (6 up-regulated and 4 down-regulated and 4 down-regulated genes) in all genes) six GEO in all six datasets. GEO datasets.

3.2. The Hub Genes Expressions Are Associated with Clinical Prognosis of the Hub Cancer Patients Our differential expression analysis suggested that the expression levels of the hub genes vary with the molecular subtypes of the breast, lung, and stomach cancers and, Cancers 2021, 13, x 9 of 23

Cancers 2021, 13, 3847 8 of 20 3.2. The Hub Genes Expressions Are Associated with Clinical Prognosis of the Hub Cancer Patients Our differential expression analysis suggested that the expression levels of the hub in mostgenes cases,vary with elevates the molecular with increased subtypes tumor of the breast, stages lung, of the and six stomach cancer cancers hub (Figure and, in3A–C, Figuremost S1). cases, Notably, elevates we with found increased that thetumor six st hubages genes of the weresix cancer significantly hub (Figure (all 3A–C,p-value Fig- < 0.05) associatedure S1). Notably, with shorter we found survival that the duration six hub of genes the breastwere significantly cancer cohorts (all p while-value none< 0.05) of the genesassociated was significantly with shorter associated survival duration (all p > 0.05)of the with breast a prognosiscancer cohorts of prostate while none cancer of the cohorts (Figuregenes3 D).was Furthermore, significantly associated with the (all exception p > 0.05) with of TNPO2a prognosisin lungof prostate cancer, cancerOBSCN cohortsin liver cancer,(Figure and 3D). SEC16 Furthermore, in gastric with and the colon exception cancers, of the TNPO2 high in expression lung cancer, levels OBSCN of the in hublivergenes predictedcancer, aand worse SEC16 prognosis in gastric ofand the colon patients. cancers, Similarly, the high expression we queried levels the of survival the hub differencesgenes betweenpredicted the a mRNA worse prognosis expression of levelsthe patients. of each Simi hublarly, gene we acrossqueried the thecombined survival differences cohortsof the hubbetween cancers the and mRNA found expression that higher levels mRNA of each hub expression gene across levels the combined of RAB31 cohorts, IRAK3 of, theSEC16B , hub cancers and found that higher mRNA expression levels of RAB31, IRAK3, SEC16B, and LIN9 predicted shorter survival durations of the hub cancer cohorts. However, our and LIN9 predicted shorter survival durations of the hub cancer cohorts. However, our resultsresults predicted predicted that that the the mRNA mRNA expression levels levels of of SEC16BSEC16B andand TNPO2TNPO2 had hadno survival no survival significancesignificance to to the the cohorts cohorts (Figure (Figure S2).

FigureFigure 3. The 3. cancer The cancer hub gene hub signatures gene signatures are of clinicalare of clinical prognostic prognostic relevance. relevance. (A) Heatmap (A) Heatmap and ( Band) trend (B) trend plot plot showing the differentialshowing the expression differential levels expression of the levels hub genesof the betweenhub genes tumor between stages tumor across stages the across six the cancers. six cancers. (C) Bubble (C) Bubble plot plot showing significantshowing differences significant in differences the expression in the levels expression of the levels hub of genes the hub in thegenes molecular in the mo subtypeslecular subtypes of the of breast, the breast, lung, lung, and and stomach stomach cancers. (D) Kaplan–Meir plot showing survival differences between hub cancers cohorts with high and low gene cancers. (D) Kaplan–Meir plot showing survival differences between hub cancers cohorts with high and low gene expression expression levels. * p < 0.05; ** p < 0.01; *** p < 0.001. HR: Hazard ratio, ns: non-significant. levels. * p < 0.05; ** p < 0.01; *** p < 0.001. HR: Hazard ratio, ns: non-significant. 3.3. Gene–Gene and Protein–Proteins Interactions, and Functional Enrichment of the Hub 3.3.Genes Gene–Gene and Protein–Proteins Interactions, and Functional Enrichment of the Hub Genes ToTo predict predict the the biological biological rolesroles of the the identified identified DEGs, DEGs, we we conducted conducted GO and GO KEGG and KEGG pathwaypathway enrichment enrichment analyses. analyses. In terms of of the the KEGG KEGG pathways, pathways, the theup-regulated up-regulated genes genes werewere enriched enriched in in cellular cellular senescencesenescence and and neurotrophin neurotrophin signaling signaling pathways pathways (Figure (Figure 4A). 4A). These DEGs were, however, significantly enriched in multiple biological processes related These DEGs were, however, significantly enriched in multiple biological processes related to cellular protein trafficking, regulation of cellular immune responses, and immune system evasion (Figure4B and Table S1). Our KEGG pathway enrichment analysis predicted that the down-regulated genes mainly participated in longevity-associated signaling pathways, including autophagy, AMPK, and insulin signaling pathways, while the enriched biological functions of the downregulated DEGs were autophagy, exocrine system development, and protein ubiquitination. Analysis of gene–gene interaction suggested that these six gene sets form a gene co-expression network interaction with known oncogenes including TRIM41, RHOQ, ZNF746, SBSPON, TRIR, AFMID, APOB, CAPN14, HSPB9, AURKB, PLPPR3, CDR1, Cancers 2021, 13, 3847 9 of 20

ZNF683, SLC43A3, WRN, PSD, CPAMD8, BCL6, and FAM24OC (Figure4C), while the PPI network of up-regulated DEGs consists of 66 nodes and 284 edges with a mean node degree of 8.60 (Figure4D). Our further analysis of gene and pathway interaction suggested

Cancers 2021a, 13 high, x probability of inhibition of DNA damage response and apoptosis while11 activatingof 23 RAS/MAPK, RTK, EMT, and PI3K/AKT signaling by the hub genes (Figure S3).

FigureFigure 4. Enrichment 4. and interaction analysis of the DEGs. (A) Bubble plot of KEGG pathwayA enrichment analysis of DEGs. (B) Bubble plotEnrichment of GO biological and enrichment interaction analysis analysis of DEGs.of GO the and DEGs.pathways (enrichment) Bubble is ranked plot by of their KEGG p- pathway value.enrichment KEGG, Kyoto analysisEncyclopedia of of DEGs. Genes and (B )Genomes. Bubble ( plotC) Cycle of plot GO of biological the DEGs’ (C enrichment) gene–gene and analysis (D) protein– of DEGs. GO proteinand interaction pathways network enrichment (PPI). Node sizes is ranked are based by on theirthe degreep-value. of connectivity KEGG, of Kyotothe nodes. Encyclopedia (E) Immunofluores- of Genes and cence stain of the subcellular distribution of the hub genes within the nucleus and ER of cancer cells. Gene localizations wereGenomes. detected based (C )on Cycle immunofluorescence plot of the DEGs’ microscopy (C) of gene–gene HPA database and using (D) the protein–protein following antibodies: interaction IRAK3 network (#HPA043097),(PPI). Node OBSCN sizes (#HPA021186), are based LIN9 on (#HPA030241), the degree TNPO2 of connectivity (#HPA071498), ofSEC16B the (#HPA054292). nodes. (E) Immunoﬂuorescence stain of the subcellular distribution of the hub genes within the nucleus and ER of cancer cells. Gene localizations were detected based on immunoﬂuorescence microscopy of HPA database using the following antibodies: IRAK3 (#HPA043097), OBSCN (#HPA021186), LIN9 (#HPA030241), TNPO2 (#HPA071498), SEC16B (#HPA054292).

To further characterize the hub genes, we acquired the experimental evidence about the sub-localization of RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B in human cell Cancers 2021, 13, 3847 10 of 20

lines via the Human Protein Atlas database. Our results suggested that IRAK3 and OBSCN are localized to the vesicle and plasma membrane, respectively, SEC16B is localized to ER and plasma membrane, and LIN9 and TNPO2 are localized to the nucleoplasm (Figure4E).

3.4. Distribution and Effect of Genetic and Epigenetic Alteration of the Hub Genes in the Hub Cancers Our analysis of the frequencies of genetic alterations of the hub genes from the TCGA cancer cohorts suggested that the genetic alterations of the hub genes occur at frequencies of 30.77%, 23.27%, 21.41%, 20.88%, 20.57%, and 8.5% in stomach, lung, liver, colorectal, breast, and prostate cancer cohorts, respectively (Figure5A). Specifically, single nucleotide variation (SNV) of OBSCN is the most frequently predicted gene alteration, while SNV of SEC16B, IRAK3, TNPO2, LIN9, and RAB31 constituted 9%, 5%, 5%, 4%, and 2% of genetic alterations in the TCGA cohort of the hub cancers (Figure5B–D). In addition, our results also suggested that the missense mutation is the most frequent effective mutation of the hub genes, having the occurrence counts of >70%, while other effective mutations including the nonsense, frame_shift_ins, splice_site, frame_shift_del, in_frame_del, and in_frame_ins constituted a smaller proportion of the SNV of the gene hub in the TCGA samples of the hub cancers (Figure4D, Table S1). The majority of the mutations were predicted to be C > T and T > C transitioned, and C > G and C > A transversion (Figure5E). However, our survival analysis indicated that SNV of TNPO2 and IRAK3 are associated with a worse prognosis of LUAD and BRCA cohorts. OBSCN predicted a poor prognosis of LIHC and PRAD while LIN is associated with poor survival of BRCA cohorts (Figure5F). Our correlation analysis also suggested that the copy number alterations of the hub genes occur most frequently in BRCA, LIHC, and LUAD, while the least CNA occurs in PRAD. These CNA were mostly heterozygous amplification and heterozygous deletion. Our results predicted distinct types of CNA of the hub genes; OBSCN, LIN9, and SEC16B were predicted to exhibit amplification (heterozygous and homozygous) type of CNA while RAB31, TNPO2, and IRAK3 were predicted to exhibit heterozygous deletion and amplification types of CNA (Figure5G). Despite the low frequencies of CNV of the hub genes in PRAD, survival analysis predicted a shorter survival duration for those cohorts having CNV of OBSCN and LIN9 (Figure5H). CNA of RAB31 and SEC16B predicted poor prognosis of BRCA and LICH, respectively. With the exception of SEC16B in BRCA, the methylation levels of the six hub genes were in negative correlation (all cor < 0 and all p < 0.05) with the mRNA expression levels in the six cancer types (Figure5I). Surprisingly, only the methylation of SEC116B in BRCA predicted a significant (p < 0.05) poor prognosis of the cohorts (Figure5J,K). Cancers 2021, 13, 3847 11 of 20 Cancers 2021, 13, x 13 of 23

FigureFigure 5. Genetic 5. Genetic and epigeneticand epigenetic alterations alterations of of the the cancercancer hub hub genes genes across across the thecancer cancer types. types. (A) The (A bar) The plot bar showing plot showing the the distribution of genetic alteration frequency of the hub gene set across the six cancer types. (B) The waterfall plot showing distributionthe mutation of genetic distribution alteration of the frequency hub genes of in thethe hub cancers. gene set (C across) Bar plot the showing six cancer the percentage types. (B )mutation The waterfall frequency plot of showing the mutationeach of distributionthe genes across of thethe six hub cancer genes types. in the (D hub) The cancers. summary ( Cplot) Bar of the plot count showing and type the of percentage variants in mutationthe sample frequencyand of each ofgene the geneslevels. acrossPlot 1 (Variant the six cancerClassification) types. shows (D) The the summary count of each plot type of theof effective count andmutation. type ofPlot variants 2 (SNV class) in the shows sample the and gene count of each SNV class of the hub gene set in the six cancer types. Plot 3 (top mutated genes per sample) shows the top levels. Plot 1 (Variant Classification) shows the count of each type of effective mutation. Plot 2 (SNV class) shows the count mutated genes. (E) The Titv plot showing the distribution of transitions (Ti; A <-> G and C <-> T) and transversions (Tv) of eachmutations SNV class of ofthe the hub hub gene gene set in set the in hub the cancers. six cancer (F) The types. bubble Plot plot 3 (topshowing mutated the patient genes survival per sample) differences shows between the top the mutated genes. (mutantE) The and Titv wild plot type showing of the six the ge distributionne hub in the ofsix transitionscancer types. (Ti; (G) A A <->pie plot G and showing C <-> the T) proportion and transversions of different (Tv) types mutations of the hubof CNV gene of set each in of the the hub genes cancers. in each (cancer.F) The ( bubbleH) Bubble plot plot showing of the survival the patient differe survivalnce between differences each gene between CNV (amplifica- the mutant and tion, deletion, and wide type). The color from blue to red and the bubble size represent the statistical significance. The wild typeblack of outline the six border gene indicates hub in the Log-rank six cancer p < 0.05. types. (I) Bubble (G) A plot pie showing plot showing the correlation the proportion between methylation of different and types mRNA of CNV of each ofgene the genesexpression. in each The cancer. blue bubbles (H) Bubble represen plott negative of the correlations, survival difference while the betweenbubble size each represents gene CNV statistical (amplification, significance. deletion, and wide(J) type).The bubble The colorplot and from (K)blue Kaplan–Meir to red and curve the showing bubble sizethe single represent gene thesurvival statistical result significance.between cohorts The with black high outline and border low methylation levels of the hub genes; the color from blue to red correlates with the hazard ratio and the bubble size indicates Log-rank p < 0.05. (I) Bubble plot showing the correlation between methylation and mRNA gene expression. The represents statistical significance. The black outline border indicates Log-rank p < 0.05. blue bubbles represent negative correlations, while the bubble size represents statistical significance. (J) The bubble plot and (K) Kaplan–Meir curve showing3.5. the The single Hub gene Genes survival Expressions result Are between Associated cohorts with withthe Gene high Ex andpression low methylationProfiles Suggestive levels of of the hub genes; the color from blue toTumor red correlatesImmune Evasion with the hazard ratio and the bubble size represents statistical significance. The black outline border indicates Log-rankWe assessedp < 0.05.the correlation between the expression levels of the hub genes and the gene expression profiles suggestive of B cell, CD8+ T Cell, CD4+ T Cell, macrophage, neu- 3.5.trophil, The Hub and Genes dendritic Expressions cell infiltrations Are Associated across the with hub thecancers. Gene We Expression found that Profiles in all the Suggestive six of Tumor Immune Evasion We assessed the correlation between the expression levels of the hub genes and the gene expression profiles suggestive of B cell, CD8+ T Cell, CD4+ T Cell, macrophage, neutrophil, and dendritic cell infiltrations across the hub cancers. We found that in all the six cancer types analyzed, the expression levels of RAB31 and IRAK3 were positively correlated (all p-value < 0.05) with the gene expression profiles suggestive of infiltrations of all immune cell types analyzed (except B cells) (Figure6A). The degree of correction Cancers 2021Cancers, 13, 38472021, 13, x 14 of 23 12 of 20

cancer types analyzed, the expression levels of RAB31 and IRAK3 were positively corre- betweenlated (all the p expression-value < 0.05) level with oftheRAB31 gene expressionand gene profiles expression suggestive profiles of infiltrations suggestive of ofall immune infiltrationimmune werecell types weak analyzed to strong (except (r B = cells) 0.128–0.452) (Figure 6A). in The breast degree and of lungcorrection cancers, between and strong to verythe expression strong (r level = 0.38–0.70) of RAB31 and incolorectal, gene expression liver, profiles stomach, suggestive and prostateof immune cancers. infiltra- IRAK3 correlatedtion were strongly weak to withstrong the(r = gene0.128–0.452) expression in breast profile and lung suggestive cancers, and of strong the six to immunevery cell infiltrationsstrong (r = in 0.38–0.70) all the cancerin colorectal, types, liver, while stomach, a strong and correlationprostate cancers. between IRAK3 thecorrelated expression of strongly with the gene expression profile suggestive of the six immune cell infiltrations in OBSCNall the, LIN9 cancer, TNPO2 types, ,while and SEC16Ba strong correlationand the levels between of gene the expression expression of suggestive OBSCN, LIN9 of, immune infiltrationTNPO2, and was SEC16B observed and the for levels prostate of gene cancer. expressionLIN9 suggestiveand TNPO2 of immuneexpressions infiltration correlated stronglywas observed with the for gene prostate expression cancer. LIN9 profiles and TNPO2 suggestive expressions of immune correlated infiltrations strongly with in LICH, whilethe no gene significant expression association profiles suggestive was predicted of immune between infiltrations the expressionsin LICH, while of no other signifi- genes and the levelscant association of gene expression was predicted suggestive between the of immuneexpressions infiltration of other genes in LICH and the cohorts levels ( Figureof 6A , Tablegene S2). expression In all the suggestive six cancers, of immune the infiltration expression in LICH levels cohorts of the (Figure hub genes6A, Table were S2). strongly correlatedIn all the with six cancers, gene expressionthe expression profiles levels of suggestive the hub genes of were macrophage strongly correlated infiltration, with while a gene expression profiles suggestive of macrophage infiltration, while a lesser or negative lessercorrelation or negative with correlationthe levels of the with gene the expre levelsssion of profiles the gene suggestive expression of B-cell profiles infiltrations suggestive of B-cellwas infiltrations recorded. was recorded.

Figure 6.FigureThe 6. hub The genes hub genes expression expression correlate correlate with the the gene gene expression expression profiles profiles suggestive suggestive of tumor immune of tumor evasion. immune Heat evasion. map showing the correlation between the expression levels of the six hub genes with the levels of gene expression sugges- Heat maptive showing of (A) sixthe immune correlation cells (B cell, between CD8+ T the Cell, expression CD4+ T Cell, levels macrophage, of the six neutrophil hub genes and withdendritic) the levelsinfiltration, of gene and ( expressionB) suggestive of (A) six immune cells (B cell, CD8+ T Cell, CD4+ T Cell, macrophage, neutrophil and dendritic) infiltration, and (B) four immunosuppressive cell types, myeloid-derived suppressor cell (MDSCs), cancer-associated fibroblasts (CAF), tumor-associated macrophages (M2-TAM), and regulatory T cell (Treg) infiltration across the hub cancers. (C) Heat map showing the correlation between the expression levels of the six hub genes with the level of active cytotoxic lymphocyte and dysfunctional T cell phenotypes of the six hub cancer cohorts. (D) Summary heat map of the association between the gene expression levels and T cell exclusion phenotypes using data from 3 immunosuppressive cells. Z-score (T cell exclusion score); The interaction coefficient “(d)” of the gene divided by its standard error. Condition; T cell exclusion signature. Cancers 2021, 13, 3847 13 of 20

In addition, we assessed the correlation between the expression levels of the hub genes and the gene expression profiles suggestive of immunosuppressive cells (MDSCs, CAF, M2-TAM, and Treg) infiltrations. Surprisingly, we observed a strong to very strong correlation between the expression levels of RAB31 and IRAK3, and the gene expression profiles suggestive of Treg infiltration, while a very strong to excellent correlation was observed with the levels of the gene expression profiles suggestive of CAF infiltrations in the hub cancers. The expression levels of OBSCN, LIN9, TNPO2, and SEC16B correlates positively with the gene expression profiles suggestive of Treg infiltration. TNP02 and OBSCN expressions were weakly correlated, while LIN9 expression correlates negatively with the gene expression profiles suggestive of CAF infiltration in the hub cancers. SEC16B expression was weakly correlated with the gene expression profiles suggestive of CAF infiltration of BRCA and PRAD, and inversely correlated with the gene expression profiles suggestive of CAF infiltration in COAD and STAD. In sharp contrast with the correlation between the expression levels of the hub genes and the gene expression profiles suggestive of M1 macrophages infiltration, we observed that the expressions of the hub genes were inversely correlated with the gene expression profiles suggestive of M2-TAM infiltrations in the six cancer types. Similarly, the expression levels of RAB31, IRAK3, and SEC16B were inversely correlated, while TNPO2 and LIN9 were positively correlated with gene expression profiles suggestive of MDSCs infiltration in the six cancer types (Figure6B, Table S3). In order to summarize the correlation between the hub gene expression level and the gene expression profiles suggestive of the immune/immunosuppressive cell infiltrations, we queried the associations between the expression levels of these hub genes and the gene expression profile suggestive of CTL, dysfunctional T cell phenotypes, and T cell exclusion in the hub cancers. Interestingly, our results suggested that the expression levels of the hub genes were inversely correlated with the levels of gene expression suggestive of CTL infiltration in the six cancers (Figure6C). The high expression levels of RAB31, IRAK3, and TNPO2 were strongly correlated with gene expression profiles suggestive of dysfunctional T cell phenotypes in the six cancers. High expression levels of OBSCN correlated with gene expression profiles suggestive of dysfunctional T cells only in lung and breast cancers, and LIN9 only in colorectal and lung cancers, while the expression of SEC16B correlates with gene expression profiles suggestive of dysfunctional T cells in breast, lung, and colorectal cancers (Figure6C). Furthermore, our analysis suggested that the expression of RAB31 could be associated with the gene expression profiles suggestive of T cell exclusion phenotypes via cancer-associated fibroblast while the expression levels of TNPO2 and LIN9 could be associated with the gene expression profiles suggestive of T cell exclusion phenotypes via MDSC (Figure6D).

3.6. The Hub Genes Expressions Predicted Chemo- and Immune Checkpoint Blockade Therapies Resistance The results of the present study suggested a signiﬁcant correlation between the mRNA expression levels of the hub genes and IC50 values of 70 small molecules. Interestingly, high mRNA expression levels of RAB31 and SEC16B predicted cancer cell lines resistant to 62 and 17 small molecule anticancer drugs, respectively. The mRNA expression levels of OBSCN and IRAK3 were associated with resistance to vemurafenib, while the IC50 concentrations of other drugs correlate negatively with the expression levels of the hub genes in the cancer cell lines (Figure7A). In addition, we link the expression levels of the hub genes and response to anti-cancer therapy using transcriptome-level data of the cancer patients. Interestingly, our results suggested that IRAK3, RAB31, LIN9, and OBSCN has strong prognostic power with estimated AUC values of 0.695, 0.644, 0.689, and 0.79, respectively. The expression level of TNPO2 predicted week prognostic power for clinical utility (AUC = 0.546) while the expression levels of SEC16B predicted no signiﬁcant association with chemotherapy sensitivity of the hub cancers (Figure7B). Finally, we access the gene prioritization of hub genes in order to predict a generalized role of each gene’s associations with ICB response outcome and phenotypes in genetic screens (CRISPR screens). Our results suggested that Cancers 2021, 13, x 16 of 23

the hub genes and response to anti-cancer therapy using transcriptome-level data of the cancer patients. Interestingly, our results suggested that IRAK3, RAB31, LIN9, and OBSCN has strong prognostic power with estimated AUC values of 0.695, 0.644, 0.689, and 0.79, respectively. The expression level of TNPO2 predicted week prognostic power for clinical utility (AUC = 0.546) while the expression levels of SEC16B predicted no significant asso- Cancers 2021, 13, 3847 14 of 20 ciation with chemotherapy sensitivity of the hub cancers (Figure 7B). Finally, we access the gene prioritization of hub genes in order to predict a generalized role of each gene’s associations with ICB response outcome and phenotypes in genetic screens (CRISPR highscreens). gene Our expression results levelssuggested of the that hub high genes gene are expression associated levels with of resistance the hub genes to worse are patientas- outcomessociated with to PD1, resistance CTL4A, to and worse PDL1 patient immunotherapies outcomes to PD1, in the CTL4A, immune and checkpoint PDL1 immuno- blockage datasetstherapies (Figure in the7 immuneC, upper checkpoint panel). The blockage prioritization datasets of the(Figure hub 7C, genes upper in the panel). gene The expression pri- proﬁlesoritization suggestive of the hub of tumorgenes in immune the gene evasion expression and profiles immunotherapy suggestive response of tumor occurs immune in the orderevasion of RAB31and immunotherapy> TNPO2 > OBSCN response> IRAK3occurs in> LIN9the order> SEC16B of RAB31(Figure > TNPO27C). > OBSCN > IRAK3 > LIN9 > SEC16B (Figure 7C).

FigureFigure 7. The 7. The cancer cancer hub hub genes genes are associatedare associated with with poor poor response response to chemotherapy to chemotherapy and immuneand immune checkpoint checkpoint blockage therapyblockage outcome. therapy (A) outcome. Bubble plot (A) of Bubble the correlation plot of the between correlation mRNA between expression mRNA expression levels of the levels hub of genes the hub and genes CTRP and drug CTRP drug sensitivity. The color from blue to red represents the correlation between mRNA expression and the IC50 sensitivity. The color from blue to red represents the correlation between mRNA expression and the IC50 value of the small value of the small molecule drugs. The positive correlation means that the gene’s high expression is resistant to the drug, moleculeand vice drugs. versa. The The positive bubble size correlation positively means correlates that the with gene’s the FDR high significance. expression The is resistant black outline to the border drug, andindicates vice versa.FDR < The bubble0.05. size(B) The positively ROC plotter correlates of the withassociation the FDR between signiﬁcance. the gene The expression black outline and sensitivity border indicates to chemotherapy FDR < 0.05. in ( Bthe) The hub ROC plottercancer. of the (C association) Heat map betweendepictingthe the gene association expression between and sensitivitythe hub genes to chemotherapywith lymphocyte-mediated in the hub cancer. tumor (Ckilling) Heat in map depictingCRISPR the screens association and outc betweenome in theICB hubsub cohort. geneswith lymphocyte-mediated tumor killing in CRISPR screens and outcome in ICB sub cohort.

3.7. The Cancer Hub Genes Are Potentia Druggable Targets of Ovatodiolide We analyzed the potential druggability of the hub genes by ovatodiolide, an anti- cancer phytochemical identified by our group previously [17,19]. Hence, we performed a molecular docking simulation for ovatodiolide and the cancer hub genes. Our results suggested that ovatodiolide has high possibilities of interacting with the crystal structure of the hub genes RAB31 (PDB:2fg5), IRAK3 (PDB:6ruu), OBSCN (PDB:2e01), LIN9 Cancers 2021, 13, 3847 (PDB:6c48), TNPO2 (PDB:2z5j), and SEC16B (PDB:3mzk) with estimated binding affinities15 of 20 of −6.5, −8.4, −6.2, −6.5, −8.1, and −8.5 Kcal/mol, respectively. Our analysis of the interactions between the hub genes (receptors) and the ligand (ovatodiolide) suggested that ovatodiolide interacts with the hub genes by several hydrogen bonding and alky interactions. receptorsThe receptors and ligand and ligand complexes complexes were predictedwere predicted to be to further be further stabilized stabilized by various by various Vander wallVander forces wall around forces the around ovatodiolide the ovatodiolide backbone ba withckbone the with respective the respective amino acidamino residue acid res- of the receptors.idue of the Several receptors. hydrophobic Several hydrophobic contacts were co predictivelyntacts were predictively observed between observed the between hub genes andthe ovatodiolide. hub genes and The ovatodiolide. predicted interactions The predicted and interactions binding afﬁnities and binding suggested affinities that sug-IRAK3, SEC168,gested that and IRAK3TNPO2, SEC168,were likely and TNPO2 the most were favored likely the receptors most favored for ovatodiolide receptors for (Figure ova- 8, Tabletodiolide3). (Figure 8, Table 3).

FigureFigure 8. Two 8. Two dimensional dimensional (2D) (2D) representation representation of of the the docking docking proﬁleprofile of ovatodiolide (OV) (OV) with with the the crystal crystal structures structures of of the cancer hub genes RAB31 (PDB:2fg5), IRAK3 (PDB:6ruu), OBSCN (PDB:2e01), LIN9 (PDB:6c48), TNPO2 the cancer hub genes RAB31 (PDB:2fg5), IRAK3 (PDB:6ruu), OBSCN (PDB:2e01), LIN9 (PDB:6c48), TNPO2 (PDB:2z5j), and (PDB:2z5j), and SEC16B (PDB:3mzk). Each complex represents the interaction of the best docking pose between the SEC16B (PDB:3mzk). Each complex represents the interaction of the best docking pose between the receptor and the ligand. receptor and the ligand.

Table 3. Docking proﬁle of ovatodiolide with the cancer hub genes.

Cancer Hub Genes Interactions IRAK3 LIN9 OBSCN RAB31 SEC16B TNPO2

∆G = −8.4 −6.5 −6.2 −6.5 −8.5 −8.1 (Kcal/mol) GLN88 ALA156 Conventional SER295 GLN401 (2.50 А) GLY15 (2.33 А) H-bond (2.92 А) (2.21 А) ALA12 (3.57 А) GLN106 (3.77 А) (3.34 А) ILE396 LEU264 ALA380 ALA386 TYR340 ILE34 PRO214 ARG376 π-alkyl PHE400 LEU90 ALA297 PHE30 PRO155 LEU419 LYS390 PRO109 LYS377 TYR404 Cancers 2021, 13, 3847 16 of 20

Table 3. Cont.

Cancer Hub Genes Interactions IRAK3 LIN9 OBSCN RAB31 SEC16B TNPO2 ASN298 ILE457 ASP311 ALA17 ARG495 SER293 PRO16 LYS119 TRP212 TRP460 Van der waal ILE294 ALA405 GLN15 GLY17 ALA108 ARG464 forces LEU338 VAL408 PHE11 HIS32 TRP107 ASP384 TRP339 VAL9 ASP31 TRP153 ALA423 HIS337 LYS13 ALA381 VAL373 ALA297 (3.68) ILE396 (3.77) LEU419 (3.57) Hydrophobic LEU90A PHE30 PRO109 (3.70) HIS337 (3.71) GLN401 (3.95) ALA423 (3.62) Interaction (А) (3.70) (3.80) PRO155 (3.60) TYR340 (3.80) TYR404 (3.75) TRP460 (3.58)

4. Discussion In the current study, six up-regulated DEGs were universally identified via integrated analysis of datasets of the six hub cancers. The identification of these DEGs in these hub cancers is suggestive of their concrete contribution in tumor-driving events and potential association with tumor progression and metastasis. These hub genes are likely to hold significance in the genesis of the hub cancers and might be prioritized for common biomarker detection among different cancers. The KEGG enrichment analysis suggests that the DEGs aid in cellular senescence and neurotrophin signaling pathways. The DEGs were, however, significantly enriched in multiple biological processes related to cellular protein trafficking, regulation of cellular immune responses, and immune system evasion, which are critical parameters involved in the tumor progression, metastatic progression, and treatment failure, and a major cause of patient death. Similarly, our analysis of gene and pathway interaction suggests a high probability of inhibition of DNA damage response and apoptosis while activating RAS/MAPK, RTK, EMT, and PI3K/AKT by the hub genes. RAS/MAPK, RTK, EMT, and PI3K/AKT signaling pathways are considered master regulators of normal physiological processes and their hyper-activation has been significantly correlated with growth, proliferation, metastasis, and drug resistance across various human cancers [34,38]. Furthermore, our enrichment analysis suggested that the down-regulated DEGs mainly participated in longevity-associated signaling pathways, such as autophagy, AMPK, and insulin signaling pathways. Thus, both GO and pathway enrichments suggest the involvement of the DEGs in immune invasion and migratory events of cancer development and the shorter lifespan of the cohorts. Subsequently, the six up-regulated DEGs were explored for tumor stage and subtypes expressions, and prognostic analysis, using the transcriptome data from TCGA, found that the DEGs correlated with differential expression in the molecular subtypes of the breast, lung, and stomach cancers and were significantly associated with worse overall survival of the liver, breast, lung, colorectal, and stomach cancers. Collectively, these findings suggest the significance of the hub genes as a universal biomarker signature for the six major cancer types. More importantly, our results suggest that these hub genes could serve as both early diagnostic and prognostic indicators for the six major cancer types. The tumor microenvironment consists of tumor cells, stromal cells, and the infiltrating immune cells [39]. Immunotherapy has revolutionized the treatment of cancers, but the mechanisms that regulate immunity in the TME are complex and require more investigation. The connections among these cancer hub genes and their roles within the TME, therapeutic responses, and patient survival were predicted in this study. Our results suggest that the expression levels of RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B were positively associated with the levels of the gene expression suggestive of B cell, CD8+ T Cell, CD4+ T Cell, macrophage, neutrophil, and dendritic cell infiltrations. Consistently, the gene expression profiles suggestive of immune infiltration appeared to be a predictor Cancers 2021, 13, 3847 17 of 20

of dysfunctional T cell phenotypes and a worse prognosis of the cohort of those cancers. In agreement with our findings, previous studies indicated that NK (resting and activated) cell, monocyte, resting mast cell, neutrophil, M1 and M2 TAM, activated dendritic cell, CD4+ T cell, and CD8+ T cell infiltrations were significant predictors of disease relapse, even after accounting for known prognostic indicators, including adjuvant therapy [40]. The functional differences between M1- and M2-polarized tumor-associated macrophages (TAMs) play essential roles in tumorigenesis and therapeutic development [41]. Tumors are likely to contain macrophages in any of these states. The M1 and M2 TAMs are associated with distinct immunomodulatory roles; they represent ex- tremes of a spectrum of functional states rather than truly distinct cell types [41]. The present study suggested a strong positive association between the expression levels of the hub genes and the gene expression profiles suggestive of M0 macrophages and a negative association with M2 subtypes in the six cancers. Furthermore, the hub genes expression signatures of M2 TAMs are inversely associated with gene expression profiles suggestive of T cell exclusion phenotypes (Figure6D), indicating that the hub genes are unlikely to play a role in macrophages dependent regulation of T cell exclusion phenotypes in the hub cancers. Conversely, the gene expression profiles suggestive of dysfunctional T cell phenotypes and high survival risk (Figure6C) predicted in cohorts with high expression levels of the hub genes suggest the oncogenic role of Mo macrophages in these cancers, a conclusion supported by previous studies [40,42]. We found that the expression levels of the hub genes positively correlated with the gene expression profiles suggestive of CAF and Treg infiltration in all the cancers. Previous experiments have shown that tumors attract Tregs to evasion of the anti-cancer immune response [43,44]. Our findings suggested that the six hub genes are likely to be associated with the regulation of Treg infiltration and supported the tumor-promoting role of Tregs, which facilitated T cell exclusion. CAFs inhibit T cell expansion by promoting the secretion and expression of immune checkpoint molecules and consequently impede anti-tumor immune response [45]. The strong correlation between the expression levels of RAB31, IRAK3, and OBSCN and the gene expression profile suggestive of CAF infiltrations strongly supported the tumor-promoting role of these genes in the hub cancers. However, the negative association between the levels of LIN9 and SEC16B and the gene expression profile suggestive of CAFs infiltrations of the hub cancers suggested that the hub genes are likely to be involved in the regulation of tumor immune evasion via different mechanisms. Genes highly expressed in tumor cells were expected to have positive associations with tumor purity, while those highly expressed in the tumor microenvironment are in negative associations with tumor purity [46]. Notably, RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B expressions were inversely correlated with the gene expression profiles suggestive of tumor purity of the hub cancer. Based on these findings, we proposed that RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B are mainly expressed in the TME rather than in the tumor cells; our results suggested that these cancer hub genes are associated with the regulation of infiltration of immune and immunosuppressive cells from the tumor microenvironment into the tumor tissues. In addition, our results suggested an association between high mRNA expression levels of RAB31 and SEC16B and the resistance of various cancer cell lines to 62 and 17 of small molecule anti-cancer drugs, respectively. Furthermore, we link the expression levels of the hub genes and response to anti-cancer therapy using transcriptome-level data of the cancer patients. Interestingly, the high expression levels of IRAK3, RAB31, LIN9, and OBSCN predicted resistance of the cohort to chemotherapy. Moreover, these high expression levels of the hub genes also predicted worse patient outcomes to PD1, CTL4A, and PDL1 immunotherapies in the immune checkpoint blockage datasets. Molecular docking is valuable in silico tool for modelling the possible interactions between a drug candidate and a target molecule [47], allowing for the depiction of the behavior of the drug candidate at the binding cavity of the receptor and elucidating the possible biological processes regulated by the drug candidate [48,49]. Interestingly, our Cancers 2021, 13, 3847 18 of 20

molecular docking study suggested that the ovatodiolide docked well into the binding cavity of RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B with estimated binding affinities ranging from 6.2–8.5 Kcal/mol. Non-covalent interactions such as hydrogen bonding, hydrophobic and ionic interactions, and Van der Waals forces play crucial roles in stabilizing the interaction between the drug candidate and protein targets [50]. Our docking analysis suggested that the interactions of ovatodiolide with the hub genes predominantly involved hydrogen bonds, Van der Waals forces, alkyl interaction, and various hydrophobic contacts (Figure8, Table3), having higher estimated binding affinities for the binding pockets of RAK3, SEC168, and TNPO2 than it does for other genes. The predicted Van der Waals forces around the ovatodiolide backbone with the respective amino acid of the receptors would likely create a strong cohesive environment, which could further stabilize the complexes [51]. Altogether, these molecular docking suggested that ovatodiolide has the molecular properties to interact with the binding site of RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B. However, the higher numbers of interactions and binding affinities predicted between the ovatodiolide and RAK3, SEC168, and TNPO2 suggested that the ovatodiolide would likely have higher druggable potentials against these genes than it does for OBSCN, LIN9, and IRAK3. Further in vitro and in vivo studies are ongoing in order to validate our in silico findings and to fully evaluate the therapeutic efficacy of this compound against the hub genes and cancers.

5. Conclusions Collectively, our study suggested that the cancer hub genes, namely, RAB31, IRAK3, OBSCN, LIN9, TNPO2, and SEC16B, are crucial biomarkers of the immuno-oncology context of the tumor microenvironment, tumor staging, prognosis, and therapy response in liver, lung, stomach prostate, and colorectal cancer. Our results suggested that the cancer hub genes were associated with the suppressive nature of the TME and the poor prognosis of the patients. This cancer hub gene signature may serve as a valuable theranostic signature in the clinical setting upon further validation. On the translational forefront, our ﬁndings suggested that ovatodiolide could be an immunotherapeutic agent for the druggability of these hub genes.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/cancers13153847/s1. Table S1: Mutation frequency of the hub genes in the hub cancers, Table S2: Correlation of gene expression and tumor infiltration of the immune cells in hub cancer, Table S3: Correlation of gene expression and tumor infiltration of the immunosuppressive cells, Figure S1: Summary the associations between subtypes and gene expression, Figure S2: Kaplan–Meir plots survival differences between the mRNA expression levels of the hub genes in the combined cohorts of the hub cancer, Figure S3: Interaction map of the hub genes and pathway. Author Contributions: J.-H.C., B.L. and D.T.W.T. collected data and wrote the manuscript; A.T.H.W., J.-C.L. and C.-L.H. performed data analyses and prepared manuscript; T.-Y.C. designed and oversaw the study. All authors have read and agreed to the published version of the manuscript. Funding: A.T.H. Wu is supported by the research grants provided by Taipei Medical University and Ministry of Education, Taiwan (DP2-110-21121-03-C-09 and DP2-110-21121-01-H-03-03). J.-C. Lee and C.-L. Ho are funded by Tri-Service General Hospital research grants (TSGH-D-110079) and (TSGH-C03-110022), respectively. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: The datasets generated and analyzed in this study will be available upon reasonable request. Conflicts of Interest: The authors declare no conflict of interest. Cancers 2021, 13, 3847 19 of 20

References 1. Sung, H.; Ferlay, J.; Siegel, R.L.; Laversanne, M.; Soerjomataram, I.; Jemal, A.; Bray, F. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. 2021, 71, 209–249. [CrossRef] 2. Siegel, R.L.; Miller, K.D.; Jemal, A. Cancer statistics, 2019. CA Cancer J. Clin 2019, 69, 7–34. [CrossRef] 3. Bray, F.; Ferlay, J.; Soerjomataram, I.; Siegel, R.L.; Torre, L.A.; Jemal, A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin. 2018, 68, 394–424. [CrossRef] 4. Iorio, M.V.; Croce, C.M. MicroRNA dysregulation in cancer: Diagnostics, monitoring and therapeutics. A comprehensive review. EMBO Mol. Med. 2012, 4, 143–159. [CrossRef] 5. Yadav, M.; Pradhan, D.; Singh, R.P. Integrated analysis and identification of nine-gene signature associated to oral squamous cell carcinoma pathogenesis. 3 Biotech 2021, 11, 215. [CrossRef] 6. Zhang, J.-J.; Hong, J.; Ma, Y.-S.; Shi, Y.; Zhang, D.-D.; Yang, X.-L.; Jia, C.-Y.; Yin, Y.-Z.; Jiang, G.-X.; Fu, D.; et al. Identified GNGT1 and NMU as Combined Diagnosis Biomarker of Non-Small-Cell Lung Cancer Utilizing Bioinformatics and Logistic Regression. Dis. Markers 2021, 2021, 6696198. [CrossRef] 7. Yan, X.; Liu, X.-P.; Guo, Z.-X.; Liu, T.-Z.; Li, S. Identification of Hub Genes Associated with Progression and Prognosis in Patients With Bladder Cancer. Front. Genet. 2019, 10.[CrossRef] 8. Khedkar, H.N.; Wang, Y.-C.; Yadav, V.K.; Srivastava, P.; Lawal, B.; Mokgautsi, N.; Sumitra, M.R.; Wu, A.T.H.; Huang, H.-S. In-Silico Evaluation of Genetic Alterations in Ovarian Carcinoma and Therapeutic Efficacy of NSC777201, as a Novel Multi-Target Agent for TTK, NEK2, and CDK1. Int. J. Mol. Sci. 2021, 22, 5895. [CrossRef][PubMed] 9. Ho, M.Y.; Mackey, J.R. Presentation and management of docetaxel-related adverse effects in patients with breast cancer. Cancer Manag. Res. 2014, 6, 253. [CrossRef] 10. Demaria, M.; O’Leary, M.N.; Chang, J.; Shao, L.; Liu, S.; Alimirah, F.; Koenig, K.; Le, C.; Mitin, N.; Deal, A.M. Cellular senescence promotes adverse effects of chemotherapy and cancer relapse. Cancer Discov. 2017, 7, 165–176. [CrossRef] 11. Choudhari, A.S.; Mandave, P.C.; Deshpande, M.; Ranjekar, P.; Prakash, O. Corrigendum: Phytochemicals in Cancer Treatment: From Preclinical Studies to Clinical Practice. Front. Pharmacol. 2020, 11, 175. [CrossRef] 12. Chen, Y.-L.; Lan, Y.-H.; Hsieh, P.-W.; Wu, C.-C.; Chen, S.-L.; Yen, C.-T.; Chang, F.-R.; Hung, W.-C.; Wu, Y.-C. Bioactive cembrane diterpenoids of Anisomeles indica. J. Nat. Prod. 2008, 71, 1207–1212. [CrossRef] 13. Chiu, N.Y.; Chang, K.H. The Illustrated Medicinal Plants of Taiwan; SMC Publisher: Taipei, Taiwan, 1992. 14. Momose, Y.; Nimura, M.; Arisawa, M.; Hayashi, T.; Takeda, R.; Nakanishi, S. Hypotensive activity of ovatodiolides isolated from a Chinese crude drug ‘Fang Feng Cao’. Phytother. Res. 1994, 8, 482–484. [CrossRef] 15. Bamodu, O.A.; Huang, W.-C.; Tzeng, D.T.W.; Wu, A.; Wang, L.S.; Yeh, C.-T.; Chao, T.-Y. Ovatodiolide sensitizes aggressive breast cancer cells to doxorubicin, eliminates their cancer stem cell-like phenotype, and reduces doxorubicin-associated toxicity. Cancer Lett. 2015, 364, 125–134. [CrossRef][PubMed] 16. Su, Y.-k.; Bamodu, O.A.; Tzeng, Y.-M.; Hsiao, M.; Yeh, C.-T.; Lin, C.-M. Ovatodiolide inhibits the oncogenicity and cancer stem cell-like phenotype of glioblastoma cells, as well as potentiate the anticancer effect of temozolomide. Phytomedicine 2019, 61, 152840. [CrossRef] 17. Chen, J.-H.; Wu, A.T.H.; Bamodu, O.A.; Yadav, V.K.; Chao, T.-Y.; Tzeng, Y.-M.; Mukhopadhyay, D.; Hsiao, M.; Lee, J.-C. Ovato- diolide Suppresses Oral Cancer Malignancy by Down-Regulating Exosomal Mir-21/STAT3/β-Catenin Cargo and Preventing Oncogenic Transformation of Normal Gingival Fibroblasts. Cancers 2020, 12, 56. [CrossRef][PubMed] 18. Lin, C.-S.; Bamodu, O.A.; Kuo, K.-T.; Huang, C.-M.; Liu, S.-C.; Wang, C.-H.; Tzeng, Y.-M.; Chao, T.-Y.; Yeh, C.-T. Investigation of ovatodiolide, a macrocyclic diterpenoid, as a potential inhibitor of oral cancer stem-like cells properties via the inhibition of the JAK2/STAT3/JARID1B signal circuit. Phytomedicine 2018, 46, 93–103. [CrossRef][PubMed] 19. Huang, Y.-J.; Yang, C.-K.; Wei, P.-L.; Huynh, T.-T.; Whang-Peng, J.; Meng, T.-C.; Hsiao, M.; Tzeng, Y.-M.; Wu, A.T.; Yen, Y. Ovatodiolide suppresses colon tumorigenesis and prevents polarization of M2 tumor-associated macrophages through YAP oncogenic pathways. J. Hematol. Oncol. 2017, 10, 1–17. [CrossRef] 20. Liu, S.-C.; Huang, C.-M.; Chang, Y.-L.; Bamodu, O.A.; Yeh, C.-T.; Wang, H.-W.; Lee, F.-P.; Lin, C.-S. Ovatodiolide suppresses inflammatory response in BEAS-2B cells by regulating the CREB/AQP5 pathway, and sensitizes nasopharyngeal carcinoma cells to radiation therapy. Eur. J. Pharmacol. 2019, 859, 172548. [CrossRef][PubMed] 21. Xiang, J.; Ding, Y.; Li, J.; Zhao, X.; Sun, Y.; Wang, D.; Wang, L.; Chen, Y. Ovatodiolides: Scalable Protection-Free Syntheses, Configuration Determination, and Biological Evaluation against Hepatic Cancer Stem Cells. Angew. Chem. 2019, 58, 10587–10590. [CrossRef][PubMed] 22. Ho, J.-Y.; Hsu, R.-J.; Wu, C.-L.; Chang, W.-L.; Cha, T.-L.; Yu, D.-S.; Yu, C.-P. Ovatodiolide Targets β-Catenin Signaling in Suppressing Tumorigenesis and Overcoming Drug Resistance in Renal Cell Carcinoma. Evid. Based Complementary Altern. Med. 2013, 2013, 161628. [CrossRef] 23. Arisawa, M.; Nimura, M.; Fujita, A.; Hayashi, T.; Morita, N.; Koshimura, S. Biological Active Macrocyclic Diterpenoids from Chinese Drug “Fáng Féng Cáo”; II1. Derivatives of Ovatodiolids and their Cytotoxity2. Planta Med. 1986, 52, 297–299. [CrossRef] 24. Liu, C.-J.; Hu, F.-F.; Xia, M.-X.; Han, L.; Zhang, Q.; Guo, A.-Y. GSCALite: A web server for gene set cancer analysis. Bioinformatics 2018, 34, 3771–3772. [CrossRef] 25. Mayakonda, A.; Koeffler, H.P. Maftools: Efficient analysis, visualization and summarization of MAF files from large-scale cohort based cancer studies. bioRxiv 2016.[CrossRef] Cancers 2021, 13, 3847 20 of 20

26. Mermel, C.H.; Schumacher, S.E.; Hill, B.; Meyerson, M.L.; Beroukhim, R.; Getz, G. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biol. 2011, 12, R41. [CrossRef] 27. Harrington, D.P.; Fleming, T.R. A class of rank test procedures for censored survival data. Biometrika 1982, 69, 553–566. [CrossRef] 28. Andersen, P.K.; Gill, R.D. Cox’s Regression Model for Counting Processes: A Large Sample Study. Ann. Stat. 1982, 10, 1100–1120. [CrossRef] 29. Li, T.; Fu, J.; Zeng, Z.; Cohen, D.; Li, J.; Chen, Q.; Li, B.; Liu, X.S. TIMER2.0 for analysis of tumor-infiltrating immune cells. Nucleic Acids Res. 2020, 48, W509–W514. [CrossRef][PubMed] 30. Fu, J.; Li, K.; Zhang, W.; Wan, C.; Zhang, J.; Jiang, P.; Liu, X.S. Large-scale public data reuse to model immunotherapy response and resistance. Genome Med. 2020, 12, 21. [CrossRef] 31. Fekete, J.T.; Gy˝orffy, B. ROCplot.org: Validating predictive biomarkers of chemotherapy/hormonal therapy/anti-HER2 therapy using transcriptomic data of 3104 breast cancer patients. Int. J. Cancer 2019, 145, 3140–3151. [CrossRef][PubMed] 32. Trott, O.; Olson, A.J. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J. Comput. Chem. 2010, 31, 455–461. [CrossRef] 33. Lee, J.-C.; Wu, A.T.H.; Chen, J.-H.; Huang, W.-Y.; Lawal, B.; Mokgautsi, N.; Huang, H.-S.; Ho, C.-L. HNC0014, a Multi-Targeted Small-Molecule, Inhibits Head and Neck Squamous Cell Carcinoma by Suppressing c-Met/STAT3/CD44/PD-L1 Oncoimmune Signature and Eliciting Antitumor Immune Responses. Cancers 2020, 12, 3759. [CrossRef][PubMed] 34. Lawal, B.; Lee, C.-Y.; Mokgautsi, N.; Sumitra, M.R.; Khedkar, H.; Wu, A.T.H.; Huang, H.-S. mTOR/EGFR/iNOS/MAP2K1/FGFR/ TGFB1 Are Druggable Candidates for N-(2,4-Difluorophenyl)-20,40-Difluoro-4-Hydroxybiphenyl-3-Carboxamide (NSC765598), With Consequent Anticancer Implications. Front. Oncol. 2021, 11.[CrossRef][PubMed] 35. Hanwell, M.D.; Curtis, D.E.; Lonie, D.C.; Vandermeersch, T.; Zurek, E.; Hutchison, G.R. Avogadro: An advanced semantic chemical editor, visualization, and analysis platform. J. Cheminformatics 2012, 4, 17. [CrossRef][PubMed] 36. Lawal, B.; Liu, Y.-L.; Mokgautsi, N.; Khedkar, H.; Sumitra, M.R.; Wu, A.T.; Huang, H.-S. Pharmacoinformatics and preclinical studies of nsc765690 and nsc765599, potential stat3/cdk2/4/6 inhibitors with antitumor activities against nci60 human tumor cell lines. Biomedicines 2021, 9, 92. [CrossRef][PubMed] 37. Visualizer, D.S. BIOVIA, Dassault Systèmes, BIOVIA Workbook, Release 2020; BIOVIA Pipeline Pilot, Release 2020; Dassault Systèmes: San Diego, CA, USA, 2020. 38. Yang, J.; Nie, J.; Ma, X.; Wei, Y.; Peng, Y.; Wei, X. Targeting PI3K in cancer: Mechanisms and advances in clinical trials. Mol. Cancer 2019, 18, 26. [CrossRef][PubMed] 39. Lawal, B.; Lin, L.-C.; Lee, J.-C.; Chen, J.-H.; Bekaii-Saab, T.S.; Wu, A.T.; Ho, C.-L. Multi-Omics Data Analysis of Gene Expressions and Alterations, Cancer-Associated Fibroblast and Immune Infiltrations, Reveals the Onco-Immune Prognostic Relevance of STAT3/CDK2/4/6 in Human Malignancies. Cancers 2021, 13, 954. [CrossRef] 40. Kamal, Y.; Dwan, D.; Hoehn, H.J.; Sanz-Pamplona, R.; Alonso, M.H.; Moreno, V.; Cheng, C.; Schell, M.J.; Kim, Y.; Felder, S.I.; et al. Tumor immune infiltration estimated from gene expression profiles predicts colorectal cancer relapse. OncoImmunology 2021, 10, 1862529. [CrossRef] 41. Noy, R.; Pollard, J.W. Tumor-associated macrophages: From mechanisms to therapy. Immunity 2014, 41, 49–61. [CrossRef] 42. Ali, H.R.; Chlon, L.; Pharoah, P.D.P.; Markowetz, F.; Caldas, C. Patterns of Immune Infiltration in Breast Cancer and Their Clinical Implications: A Gene-Expression-Based Retrospective Study. PLoS Med. 2016, 13, e1002194. [CrossRef] 43. Shimizu, J.; Yamazaki, S.; Sakaguchi, S. Induction of tumor immunity by removing CD25+CD4+ T cells: A common basis between tumor immunity and autoimmunity. J. Immunol. 1999, 163, 5211–5218. 44. Onizuka, S.; Tawara, I.; Shimizu, J.; Sakaguchi, S.; Fujita, T.; Nakayama, E. Tumor rejection by in vivo administration of anti-CD25 (interleukin-2 receptor alpha) monoclonal antibody. Cancer Res. 1999, 59, 3128–3133. 45. Harryvan, T.J.; Verdegaal, E.M.E.; Hardwick, J.C.H.; Hawinkels, L.J.A.C.; van der Burg, S.H. Targeting of the Cancer-Associated Fibroblast—T-Cell Axis in Solid Malignancies. J. Clin. Med. 2019, 8, 1989. [CrossRef][PubMed] 46. Li, T.; Fan, J.; Wang, B.; Traugh, N.; Chen, Q.; Liu, J.S.; Li, B.; Liu, X.S. TIMER: A Web Server for Comprehensive Analysis of Tumor-Infiltrating Immune Cells. Cancer Res. 2017, 77, e108–e110. [CrossRef] 47. Kitchen, D.B.; Decornez, H.; Furr, J.R.; Bajorath, J. Docking and scoring in virtual screening for drug discovery: Methods and applications. Nat. Rev. Drug Discov. 2004, 3, 935–949. [CrossRef][PubMed] 48. Meng, X.-Y.; Zhang, H.-X.; Mezei, M.; Cui, M. Molecular docking: A powerful approach for structure-based drug discovery. Curr. Comput. Aided Drug Des. 2011, 7, 146–157. [CrossRef] 49. Jorgensen, W.L. The many roles of computation in drug discovery. Science 2004, 303, 1813–1818. [CrossRef][PubMed] 50. Zhao, H.; Huang, D. Hydrogen bonding penalty upon ligand binding. PLoS ONE 2011, 6, e19923. [CrossRef] 51. Arthur, D.E.; Uzairu, A. Molecular docking studies on the interaction of NCI anticancer analogues with human Phosphatidylinos- itol 4,5-bisphosphate 3-kinase catalytic subunit. J. King Saud Univ. Sci. 2019, 31, 1151–1166. [CrossRef]